Process Mining

Process Mining

Bachelor Thesis, Master Thesis

Process Mining is a family of techniques combining data science and business process management to support the analysis of (business) processes solely based on event logs, typically recorded by enterprise information systems. These event logs contain events about what people, machines, and organizations are really doing. By using data mining, machine learning, and other related data analysis techniques, novel insights can be obtained to address performance and compliance problems of processes.

Process Mining - Process Map

In process mining there exist mainly three categories of analysis:

  • (1) process discovery, which deals with the reconstruction of a process model from an event log,
  • (2) conformance checking, which checks if the reality conforms to the desired way of executing the process, and
  • (3) enhancement, which enriches process models with additional information.

All three analysis categories support an analyst finding issues in a process using data-centric techniques.

We offer various topics around the research area of process mining, including but not limited to:

  • Process learning: using machine learning models (i.e., deep learning) to better capture how a process has been really executed.
  • Automatic analysis recommendations: analyzing an event log such that the system provides potential insights to the analyst automatically for further (manual) analysis.
  • Visual exploration: extending visual data analysis techniques to allow better and high-quality insights extraction from large data sets.
  • Task mining: grouping low level activities that people, machines, or organizations execute within a process to allow detailed analysis of the specific tasks.

Feel free to contact us, if you are interested in one of the topic or in process mining in general for a bachelor / master thesis.

Requirements

  • Good programming skills
  • (depends) data mining, machine learning, deep learning
  • Interest and enthusiasm in the analysis of event data from processes and algorithmic challenges